FROM Transact-SQL

In Transact-SQL, the FROM clause is available on the following statements:

• DELETE
• UPDATE
• SELECT

The FROM clause is usually required on the SELECT statement. The exception is when no table columns are listed, and the only items listed are literals or variables or arithmetic expressions.

This article also discusses the following keywords that can be used on the FROM clause:

• JOIN
• APPLY
• PIVOT

Syntax for SQL Server and Azure SQL Database 

[ FROM { <table_source> } [ ,…n ] ]  

<table_source> ::=  

{ 

    table_or_view_name [ [ AS ] table_alias ]  

        [ <tablesample_clause> ]  

        [ WITH ( < table_hint > [ [ , ]…n ] ) ]  

    | rowset_function [ [ AS ] table_alias ]  

        [ ( bulk_column_alias [ ,…n ] ) ]  

    | user_defined_function [ [ AS ] table_alias ] 

    | OPENXML <openxml_clause>  

    | derived_table [ [ AS ] table_alias ] [ ( column_alias [ ,…n ] ) ]  

    | <joined_table>  

    | <pivoted_table>  

    | <unpivoted_table> 

    | @variable [ [ AS ] table_alias ] 

    | @variable.function_call ( expression [ ,…n ] )  

        [ [ AS ] table_alias ] [ (column_alias [ ,…n ] ) ] 

    | FOR SYSTEM_TIME <system_time>  

} 

<tablesample_clause> ::= 

    TABLESAMPLE [SYSTEM] ( sample_number [ PERCENT | ROWS ] )  

        [ REPEATABLE ( repeat_seed ) ]  

<joined_table> ::=  

{ 

    <table_source> <join_type> <table_source> ON <search_condition>  

    | <table_source> CROSS JOIN <table_source>  

    | left_table_source { CROSS | OUTER } APPLY right_table_source  

    | [ ( ] <joined_table> [ ) ]  

} 

<join_type> ::=  

    [ { INNER | { { LEFT | RIGHT | FULL } [ OUTER ] } } [ <join_hint> ] ] 

    JOIN 

<pivoted_table> ::= 

    table_source PIVOT <pivot_clause> [ [ AS ] table_alias ] 

<pivot_clause> ::= 

        ( aggregate_function ( value_column [ [ , ]…n ])  

        FOR pivot_column  

        IN ( <column_list> )  

    )  

<unpivoted_table> ::= 

    table_source UNPIVOT <unpivot_clause> [ [ AS ] table_alias ] 

<unpivot_clause> ::= 

    ( value_column FOR pivot_column IN ( <column_list> ) )  

<column_list> ::= 

    column_name [ ,…n ]  

<system_time> ::= 

{ 

       AS OF <date_time> 

    |  FROM <start_date_time> TO <end_date_time> 

    |  BETWEEN <start_date_time> AND <end_date_time> 

    |  CONTAINED IN (<start_date_time> , <end_date_time>)  

    |  ALL 

} 

    <date_time>::= 

        <date_time_literal> | @date_time_variable 

    <start_date_time>::= 

        <date_time_literal> | @date_time_variable 

    <end_date_time>::= 

        <date_time_literal> | @date_time_variable

Syntax for Azure SQL Data Warehouse and Parallel Data Warehouse 

FROM { <table_source> [ ,…n ] } 

<table_source> ::=  

{ 

    [ database_name . [ schema_name ] . | schema_name . ] table_or_view_name [ AS ] table_or_view_alias

    [<tablesample_clause>] 

    | derived_table [ AS ] table_alias [ ( column_alias [ ,…n ] ) ] 

    | <joined_table> 

} 

<tablesample_clause> ::=

    TABLESAMPLE ( sample_number [ PERCENT ] ) — SQL Data Warehouse only 

<joined_table> ::=  

{ 

    <table_source> <join_type> <table_source> ON search_condition  

    | <table_source> CROSS JOIN <table_source>

    | left_table_source { CROSS | OUTER } APPLY right_table_source  

    | [ ( ] <joined_table> [ ) ]  

} 

<join_type> ::=  

    [ INNER ] [ <join hint> ] JOIN 

    | LEFT  [ OUTER ] JOIN 

    | RIGHT [ OUTER ] JOIN 

    | FULL  [ OUTER ] JOIN 

<join_hint> ::=  

    REDUCE 

    | REPLICATE 

    | REDISTRIBUTE

Arguments

<table source>

Query performance may suffer with lots of tables referenced in a query. Compilation and optimization time is also affected by additional factors. These include the presence of indexes and indexed views on each and the size of the in the SELECT statement.

The order of table sources after the FROM keyword does not affect the result set that is returned. SQL Server returns errors when duplicate names appear in the FROM clause.

table_or_view_name
Is the name of a table or view.

If the table or view exists in another database on the same instance of SQL Server, use a fully qualified name in the form

database.schema.object_name.

If the table or view exists outside the instance of SQL Serverl, use a four-part name in the form linked_server.catalog.schema.object. For more information, see sp_addlinkedserver (Transact-SQL). A four-part name that is constructed by using the OPENDATASOURCE function as the server part of the name can also be used to specify the remote table source. When OPENDATASOURCE is specified, database_name and schema_name may not apply to all data sources and is subject to the capabilities of the OLE DB provider that accesses the remote object.

[AS] table_alias
Is an alias for table_source that can be used either for convenience or to distinguish a table or view in a self-join or subquery. An alias is frequently a shortened table name used to refer to specific columns of the tables in a join. If the same column name exists in more than one table in the join, SQL Server requires that the column name be qualified by a table name, view name, or alias. The table name cannot be used if an alias is defined.

When a derived table, rowset or table-valued function, or operator clause (such as PIVOT or UNPIVOT) is used, the required table_alias at the end of the clause is the associated table name for all columns, including grouping columns, returned.

WITH ( )
Specifies that the query optimizer use an optimization or locking strategy with this table and for this statement. For more information, see Table Hints (Transact-SQL).

rowset_function

Applies to: SQL Server 2008 through SQL Server 2017 and SQL Database.

Specifies one of the rowset functions, such as OPENROWSET, that returns an object that can be used instead of a table reference. For more information about a list of rowset functions, see Rowset Functions (Transact-SQL).

Using the OPENROWSET and OPENQUERY functions to specify a remote object depends on the capabilities of the OLE DB provider that accesses the object.

bulk_column_alias

Applies to: SQL Server 2008 through SQL Server 2017 and SQL Database.

Is an optional alias to replace a column name in the result set. Column aliases are allowed only in SELECT statements that use the OPENROWSET function with the BULK option. When you use bulk_column_alias, specify an alias for every table column in the same order as the columns in the file.

Note: This alias overrides the NAME attribute in the COLUMN elements of an XML format file, if present.

user_defined_function
Specifies a table-valued function.

OPENXML <openxml_clause>

Applies to: SQL Server 2008 through SQL Server 2017 and SQL Database.

Provides a rowset view over an XML document.

derived_table
Is a subquery that retrieves rows from the database. derived_table is used as input to the outer query.

derived _table can use the Transact-SQL table value constructor feature to specify multiple rows. For example, SELECT * FROM (VALUES (1, 2), (3, 4), (5, 6), (7, 8), (9, 10) ) AS MyTable(a, b);. For more information, see Table Value Constructor (Transact-SQL).

column_alias
Is an optional alias to replace a column name in the result set of the derived table. Include one column alias for each column in the select list, and enclose the complete list of column aliases in parentheses.

table_or_view_name FOR SYSTEM_TIME

Applies to: SQL Server 2016 (13.x) through SQL Server 2017 and SQL Database.

Specifies that a specific version of data is returned from the specified temporal table and its linked system-versioned history table

Tablesample clause
Applies to: SQL Server, SQL Database

Specifies that a sample of data from the table is returned. The sample may be approximate. This clause can be used on any primary or joined table in a SELECT or UPDATE statement. TABLESAMPLE cannot be specified with views.

Note: When you use TABLESAMPLE against databases that are upgraded to SQL Server, the compatibility level of the database is set to 110 or higher, PIVOT is not allowed in a recursive common table expression (CTE) query. For more information, see ALTER DATABASE Compatibility Level (Transact-SQL).

SYSTEM
Is an implementation-dependent sampling method specified by ISO standards. In SQL Server, this is the only sampling method available and is applied by default. SYSTEM applies a page-based sampling method in which a random set of pages from the table is chosen for the sample, and all the rows on those pages are returned as the sample subset.

sample_number
Is an exact or approximate constant numeric expression that represents the percent or number of rows. When specified with PERCENT, sample_number is implicitly converted to a float value; otherwise, it is converted to bigint. PERCENT is the default.

PERCENT
Specifies that a sample_number percent of the rows of the table should be retrieved from the table. When PERCENT is specified, SQL Server returns an approximate of the percent specified. When PERCENT is specified the sample_number expression must evaluate to a value from 0 to 100.

ROWS
Specifies that approximately sample_number of rows will be retrieved. When ROWS is specified, SQL Server returns an approximation of the number of rows specified. When ROWS is specified, the sample_number expression must evaluate to an integer value greater than zero.

REPEATABLE
Indicates that the selected sample can be returned again. When specified with the same repeat_seed value, SQL Server will return the same subset of rows as long as no changes have been made to any rows in the table. When specified with a different repeat_seed value, SQL Server will likely return some different sample of the rows in the table. The following actions to the table are considered changes: insert, update, delete, index rebuild or defragmentation, and database restore or attach.

repeat_seed
Is a constant integer expression used by SQL Server to generate a random number. repeat_seed is bigint. If repeat_seed is not specified, SQL Server assigns a value at random. For a specific repeat_seed value, the sampling result is always the same if no changes have been applied to the table. The repeat_seed expression must evaluate to an integer greater than zero.

Tablesample clause
Applies to: SQL Data Warehouse

Specifies that a sample of data from the table is returned. The sample may be approximate. This clause can be used on any primary or joined table in a SELECT or UPDATE statement. TABLESAMPLE cannot be specified with views.

PERCENT
Specifies that a sample_number percent of the rows of the table should be retrieved from the table. When PERCENT is specified, SQL Data Warehouse returns an approximate of the percent specified. When PERCENT is specified, the sample_number expression must evaluate to a value from 0 to 100.

Joined table
A joined table is a result set that is the product of two or more tables. For multiple joins, use parentheses to change the natural order of the joins.

Join type
Specifies the type of join operation.

INNER
Specifies all matching pairs of rows are returned. Discards unmatched rows from both tables. When no join type is specified, this is the default.

FULL [ OUTER ]
Specifies that a row from either the left or right table that does not meet the join condition is included in the result set, and output columns that correspond to the other table are set to NULL. This is in addition to all rows typically returned by the INNER JOIN.

LEFT [ OUTER ]
Specifies that all rows from the left table not meeting the join condition are included in the result set, and output columns from the other table are set to NULL in addition to all rows returned by the inner join.

RIGHT [OUTER]
Specifies all rows from the right table not meeting the join condition are included in the result set, and output columns that correspond to the other table are set to NULL, in addition to all rows returned by the inner join.

Join hint
For SQL Server and SQL Database, specifies that the SQL Server query optimizer use one join hint, or execution algorithm, per join specified in the query FROM clause. For more information, see Join Hints (Transact-SQL).

For SQL Data Warehouse and Parallel Data Warehouse, these join hints apply to INNER joins on two distribution incompatible columns. They can improve query performance by restricting the amount of data movement that occurs during query processing. The allowable join hints for SQL Data Warehouse and Parallel Data Warehouse are as follows:

REDUCE
Reduces the number of rows to be moved for the table on the right side of the join in order to make two distribution incompatible tables compatible. The REDUCE hint is also called a semi-join hint.

REPLICATE
Causes the values in the joining column from the table on the left side of the join to be replicated to all nodes. The table on the right is joined to the replicated version of those columns.

REDISTRIBUTE
Forces two data sources to be distributed on columns specified in the JOIN clause. For a distributed table, Parallel Data Warehouse will perform a shuffle move. For a replicated table, Parallel Data Warehouse will perform a trim move. To understand these move types, see the “DMS Query Plan Operations” section in the “Understanding Query Plans” topic in the Parallel Data Warehouse product documentation. This hint can improve performance when the query plan is using a broadcast move to resolve a distribution incompatible join.

JOIN
Indicates that the specified join operation should occur between the specified table sources or views.

ON
Specifies the condition on which the join is based. The condition can specify any predicate, although columns and comparison operators are frequently used, for example:

SELECT p.ProductID, v.BusinessEntityID
FROM Production.Product AS p
JOIN Purchasing.ProductVendor AS v
ON (p.ProductID = v.ProductID);

When the condition specifies columns, the columns do not have to have the same name or same data type; however, if the data types are not the same, they must be either compatible or types that SQL Server can implicitly convert. If the data types cannot be implicitly converted, the condition must explicitly convert the data type by using the CONVERT function.

There can be predicates that involve only one of the joined tables in the ON clause. Such predicates also can be in the WHERE clause in the query. Although the placement of such predicates does not make a difference for INNER joins, they might cause a different result when OUTER joins are involved. This is because the predicates in the ON clause are applied to the table before the join, whereas the WHERE clause is semantically applied to the result of the join.

For more information about search conditions and predicates, see Search Condition (Transact-SQL).

CROSS JOIN
Specifies the cross-product of two tables. Returns the same rows as if no WHERE clause was specified in an old-style, non-SQL-92-style join.

left_table_source { CROSS | OUTER } APPLY right_table_source
Specifies that the right_table_source of the APPLY operator is evaluated against every row of the left_table_source. This functionality is useful when the right_table_source contains a table-valued function that takes column values from the left_table_source as one of its arguments.

Either CROSS or OUTER must be specified with APPLY. When CROSS is specified, no rows are produced when the right_table_source is evaluated against a specified row of the left_table_source and returns an empty result set.

When OUTER is specified, one row is produced for each row of the left_table_source even when the right_table_source evaluates against that row and returns an empty result set.

For more information, see the Remarks section.

left_table_source
Is a table source as defined in the previous argument. For more information, see the Remarks section.

right_table_source
Is a table source as defined in the previous argument. For more information, see the Remarks section.

PIVOT clause
table_source PIVOT
Specifies that the table_source is pivoted based on the pivot_column. table_source is a table or table expression. The output is a table that contains all columns of the table_source except the pivot_column and value_column. The columns of the table_source, except the pivot_column and value_column, are called the grouping columns of the pivot operator. For more information about PIVOT and UNPIVOT, see Using PIVOT and UNPIVOT.

PIVOT performs a grouping operation on the input table with regard to the grouping columns and returns one row for each group. Additionally, the output contains one column for each value specified in the column_list that appears in the pivot_column of the input_table.

aggregate_function
Is a system or user-defined aggregate function that accepts one or more inputs. The aggregate function should be invariant to null values. An aggregate function invariant to null values does not consider null values in the group while it is evaluating the aggregate value.

The COUNT(*) system aggregate function is not allowed.

value_column
Is the value column of the PIVOT operator. When used with UNPIVOT, value_column cannot be the name of an existing column in the input table_source.

FOR pivot_column
Is the pivot column of the PIVOT operator. pivot_column must be of a type implicitly or explicitly convertible to nvarchar(). This column cannot be image or rowversion.

When UNPIVOT is used, pivot_column is the name of the output column that becomes narrowed from the table_source. There cannot be an existing column in table_source with that name.

IN (column_list )
In the PIVOT clause, lists the values in the pivot_column that will become the column names of the output table. The list cannot specify any column names that already exist in the input table_source that is being pivoted.

In the UNPIVOT clause, lists the columns in table_source that will be narrowed into a single pivot_column.

table_alias
Is the alias name of the output table. pivot_table_alias must be specified.

UNPIVOT
Specifies that the input table is narrowed from multiple columns in column_list into a single column called pivot_column. For more information about PIVOT and UNPIVOT, see Using PIVOT and UNPIVOT.

AS OF <date_time>

Applies to: SQL Server 2016 (13.x) through SQL Server 2017 and SQL Database.

Returns a table with single record for each row containing the values that were actual (current) at the specified point in time in the past. Internally, a union is performed between the temporal table and its history table and the results are filtered to return the values in the row that was valid at the point in time specified by the parameter. The value for a row is deemed valid if the system_start_time_column_name value is less than or equal to the parameter value and the system_end_time_column_name value is greater than the parameter value.

FROM TO <start_date_time> TO <end_date_time>

Applies to: SQL Server 2016 (13.x) through SQL Server 2017 and SQL Database.

Returns a table with the values for all record versions that were active within the specified time range, regardless of whether they started being active before the parameter value for the FROM argument or ceased being active after the parameter value for the TO argument. Internally, a union is performed between the temporal table and its history table and the results are filtered to return the values for all row versions that were active at any time during the time range specified. Rows that became active exactly on the lower boundary defined by the FROM endpoint are included and rows that became active exactly on the upper boundary defined by the TO endpoint are not included.

BETWEEN <start_date_time> AND <end_date_time>

Applies to: SQL Server 2016 (13.x) through SQL Server 2017 and SQL Database.

Same as above in the FROM TO description, except it includes rows that became active on the upper boundary defined by the endpoint.

CONTAINED IN ( , ) (<start_date_time> , <end_date_time>)

Applies to: SQL Server 2016 (13.x) through SQL Server 2017 and SQL Database.

Returns a table with the values for all record versions that were opened and closed within the specified time range defined by the two datetime values for the CONTAINED IN argument. Rows that became active exactly on the lower boundary or ceased being active exactly on the upper boundary are included.

ALL
Returns a table with the values from all rows from both the current table and the history table.

UNION and JOIN within a FROM clause are supported within views and in derived tables and subqueries.

A self-join is a table that is joined to itself. Insert or update operations that are based on a self-join follow the order in the FROM clause.

Because SQL Server considers distribution and cardinality statistics from linked servers that provide column distribution statistics, the REMOTE join hint is not required to force evaluating a join remotely. The SQL Server query processor considers remote statistics and determines whether a remote-join strategy is appropriate. REMOTE join hint is useful for providers that do not provide column distribution statistics.

Using APPLY
Both the left and right operands of the APPLY operator are table expressions. The main difference between these operands is that the right_table_source can use a table-valued function that takes a column from the left_table_source as one of the arguments of the function. The left_table_source can include table-valued functions, but it cannot contain arguments that are columns from the right_table_source.

The APPLY operator works in the following way to produce the table source for the FROM clause:

Evaluates right_table_source against each row of the left_table_source to produce rowsets.

The values in the right_table_source depend on left_table_source. right_table_source can be represented approximately this way: TVF(left_table_source.row), where TVF is a table-valued function.

Combines the result sets that are produced for each row in the evaluation of right_table_source with the left_table_source by performing a UNION ALL operation.

The list of columns produced by the result of the APPLY operator is the set of columns from the left_table_source that is combined with the list of columns from the right_table_source.

Using PIVOT and UNPIVOT
The pivot_column and value_column are grouping columns that are used by the PIVOT operator. PIVOT follows the following process to obtain the output result set:

Performs a GROUP BY on its input_table against the grouping columns and produces one output row for each group.

The grouping columns in the output row obtain the corresponding column values for that group in the input_table.

Generates values for the columns in the column list for each output row by performing the following:

Grouping additionally the rows generated in the GROUP BY in the previous step against the pivot_column.

For each output column in the column_list, selecting a subgroup that satisfies the condition:

pivot_column = CONVERT(, ‘output_column’)

aggregate_function is evaluated against the value_column on this subgroup and its result is returned as the value of the corresponding output_column. If the subgroup is empty, SQL Server generates a null value for that output_column. If the aggregate function is COUNT and the subgroup is empty, zero (0) is returned.

Note: The column identifiers in the UNPIVOT clause follow the catalog collation. For SQL Database, the collation is always SQL_Latin1_General_CP1_CI_AS. For SQL Server partially contained databases, the collation is always Latin1_General_100_CI_AS_KS_WS_SC. If the column is combined with other columns, then a collate clause (COLLATE DATABASE_DEFAULT) is required to avoid conflicts.